← Back to community benchmarks

Qwen3.5-35B-A3B-Claude-4.6-Opus-Reasoning-Distilled

M5 Pro (20c) · 64 GB · 6bit · 2026-03-21
Performance
16k
tokens
2,013
PP tok/s
64.8
TG tok/s
8140
TTFT (ms)
29.3
Peak mem (GB)
Hardware
Chip M5 Pro (20c)
Memory 64 GB
GPU Cores 20
Software
oMLX v0.2.19
macOS macOS 26.3.1
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,400 77.6 27.3 GB view
4k 2,042 74.8 28.3 GB view
8k 2,114 71.0 28.6 GB view
16k 2,013 64.8 29.3 GB current
32k 1,730 54.2 30.6 GB view
64k 1,293 41.5 33.2 GB view
128k 823.8 25.3 38.6 GB view
195k 596.4 11.9 44.3 GB view
Batching Results
Batch Size TG tok/s Speedup
77.6 1.00×
115.6 1.49×
148.4 1.91×
174.2 2.24×